https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93067
lhyatt at gcc dot gnu.org changed:
What |Removed |Added
----------------------------------------------------------------------------
Attachment #47547|0 |1
is obsolete| |
CC| |lhyatt at gcc dot gnu.org
--- Comment #3 from lhyatt at gcc dot gnu.org ---
Created attachment 49557
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=49557&action=edit
updated patch to apply to current master
It seems I did not pursue this beyond filing this bug report with the patch.
The patch requires some minor rebasing, just due to surrounding context, so
here is a version that applies now, and still bootstraps and works. Should I
send to gcc-patches as well?
I think the main potential issue with it is that it requires input.h / input.c
to get a global variable so that frontends can let it know they are using
libcpp to read their input. (Otherwise -finput-charset is not applicable). I am
sure there is a better way to do that, but not sure the cleanest one, none of
the routines in input.c seem to make use of a diagnostic_context or similar
thing.
-Lewis
